The effect of stratified sampling of items on the estimation of test score distribution parameters by multiple matrix sampling was studied. Item difficulty and/or interitem correlations were the bases of stratification. Various item iniverses were created by computer simulation and sampled according to several plans. The results indicate that…

A subject by item matrix of test responses is shown to be a useful heuristic in criterion referenced and norm referenced test analysis, and in the teaching of measurement. The pattern of responses within the matrix provides indications of item interactions, weak deceptors, and conventional test statistics. The strong visual analogy between the…

One solution to the problem of obtaining test scores in order to assess multiple outcomes of educational programs is a basically simple technique called matrix sampling. Matrix sampling is the estimation of total test score statistics by administering random subsets of test items to randomly selected students. This chapter lays out the specific…

In discussion of assessment and its applicability to the needs of the Spanish-speaking student population, four types of assessments are presented in the context of the educationally significant testing needed in this area. The focus of this paper is on the use of comprehensive assessment to measure the repertoire of language resources that the…

Any systematic approach to the assessment of students' ability implies the use of a model. The more explicit the model is, the more its users know about what they are doing and what the consequences are. The Rasch model is a strong model where measurement is a bonus of the model itself. It is based on four ideas: (1) separation of observable…

This paper is about the structural equation modelling of quantitative measures that are obtained from a multiple facet design. A facet is simply a set consisting of a finite number of elements. It is assumed that measures are obtained by combining each element of each facet. Methods and traits are two such facets, and a multitrait-multimethod…

A central problem for the user of domain-referenced tests in instruction is deciding who has passed and who has failed. Two procedures were presented and discussed. The first, employing classical test theory, was found to be more useful for larger domains and where the passing standard is 70 percent or less. The sampling procedure suggested by…